Synaptic dynamics are a tunable substrate sculpting neural population activity

Abstract A central tenet of cerebellar computation posits that granule cells generate sparse spatiotemporal activity patterns that support precisely timed motor and cognitive outputs. Using high-speed in vivo calcium and glutamate imaging combined with slice electrophysiology, we show that heterogeneous synaptic dynamics transform mossy fiber inputs into temporally sparse, sequential patterns of GC activity. Region-specific differences in MF glutamate release shape GC response duration and sparsity, tuning the temporal statistics of population sequences to match sensorimotor associative learning demands. These findings provide direct evidence for temporally sparse GC sequences and establish synaptic dynamics as a ubiquitous, tunable mechanism structuring neural activity in time.